WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 1

Prospects and Problems of Wireless Communication for Underwater

Sensor Networkss

Lanbo Liu, Shengli Zhou, and Jun-Hong Cui

The work of L. Liu is supported by USACE AT24. The work of S. Zhou is supported by the US ONR YIP grant N00014-07-1-0805 and
the NSF grant ECCS 0725562. The work of J.-H. Cui is supported by the US NSF CAREER grant 0644190 and the US ONR YIP grant
N00014-08-1-0864.
L. Liu is with the Department of Civil and Environmental Engineering, University of Connecticut, Storrs, Connecticut 06269, (email:
lanbo@engr.uconn.edu).
S. Zhou is with the Department of Electrical and Computer Engineering, University of Connecticut, Storrs, Connecticut 06269, USA
(email: shengli@engr.uconn.edu).
J.-H. Cui is with the Department of Computer Science and Engineering, University of Connecticut, Storrs, Connecticut 06269, USA (email:
jcui@engr.uconn.edu).

Contact author: Dr. J.-H. Cui, e-mail: jcui@engr.uconn.edu; phone: 1-860-486-8951; fax: 1-860-486-4817.
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 2

Abstract

This paper reviews the physical fundamentals and engineering implementations for efficient information exchange
via wireless communication using physical waves as the carrier among nodes in an underwater sensor network. The
physical waves under discussion include sound, radio, and light. We first present the fundamental physics of different
waves; then we discuss and compare the pros and cons for adopting different communication carriers (acoustic,
radio, and optical) based on the fundamental first principles of physics and engineering practice. The discussions are
mainly targeted at underwater sensor networks with densely deployed nodes. Based on the comparison study, we
make recommendations for the selection of communication carriers for underwater sensor networks with engineering
countermeasures that can possibly enhance the communication efficiency in specified underwater environments.

Index Terms

Underwater sensor networks, wireless communication, acoustic waves, electromagnetic waves, optical waves.

I. I NTRODUCTION

In last several years, underwater sensor network (UWSN) has found an increasing use in a wide range of
applications, such as coastal surveillance systems, environmental research, autonomous underwater vehicle (AUV)
operation, to name a few [1]–[4]. By deploying a distributed and scalable sensor network in a 3-dimensional
underwater space, each underwater sensor can monitor and detect environmental parameters and events locally.
Hence, compared with remote sensing, UWSNs provide a better sensing and surveillance technology to acquire better
data to understand the spatial and temporal complexities of underwater environments. Clearly, efficient underwater
communication among units or nodes in a UWSN is one of the most fundamental and critical issues in the whole
network system design.
Present underwater communication systems involve the transmission of information in the form of sound,
electromagnetic (EM), or optical waves. Each of these techniques has advantages and limitations.
Acoustic communication is the most versatile and widely used technique in underwater environments due to
the low attenuation (signal reduction) of sound in water. This is especially true in thermally stable, deep water
settings. On the other hand, the use of acoustic waves in shallow water can be adversely affected by temperature
gradients, surface ambient noise, and multipath propagation due to reflection and refraction. The much slower speed
of acoustic propagation in water, about 1500 m/s (meters per second), compared with that of electromagnetic and
optical waves, is another limiting factor for efficient communication and networking. Nevertheless, the currently
favorable technology for underwater communication is upon acoustics.
On the front of using electromagnetic (EM) waves in radio frequencies, conventional radio does not work well in
an underwater environment due to the conducting nature of the medium, especially in the case of seawater. However,
if EM could be working underwater, even in a short distance, its much faster propagating speed is definitely a great
advantage for faster and efficient communication among nodes.
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 3

Free-space optical (FSO) waves used as wireless communication carriers are generally limited to very short
distances because the severe water absorption at the optical frequency band and strong backscatter from suspending
particles. Even the clearest water has 1000 times the attenuation of clear air, and turbid water has more than 100
times the attenuation of the densest fog. Nevertheless, underwater FSO, especially in the blue-green wavelengths,
offers a practical choice for high-bandwidth communication (10-150 Mbps, bits per second) over moderate ranges
(10-100 meters). This communication range is much needed in harbor inspection, oil-rig maintenance, and linking
submarines to land, just name a few of the demands on this front.
In this paper, we review the physical fundamentals and engineering implementations for efficient communica-
tion, via acoustic, EM and optical waves, among nodes in a UWSN. We first present the communication needs
and requirements for UWSNs in next section. Then, we discuss the fundamental physics of acoustic, radio and
optical waves, and pertinent concerns as wireless communication carriers. After that, we compare the engineering
countermeasures for the shortcomings of each individual carrier. Finally, we discuss the networking challenges for
underwater acoustic sensor networks, followed by a short summary of the applicability of three types of waves in
underwater sensor networks.

II. C OMMUNICATION R EQUIREMENTS FOR U NDERWATER S ENSOR N ETWORKS

The underwater sensor networks (UWSNs) targeted by this paper are underwater networks with densely deployed
sensor nodes. High node density is the key characteristic of such networks. Depending on the applications, we can
roughly classify the targeted dense sensor networks into two categories: 1) UWSNs for long-term non-time critical
aquatic monitoring applications (such as oceanographic data collection, pollution monitoring/detection, and off-shore
oil/gas field monitoring); 2) UWSNs for short-term time-critical aquatic exploration applications (such as submarine
detection, loss treasure discovery, and hurricane disaster recovery) [2]. The former category of UWSNs can be either
mobile or static depending on the deployment of sensor nodes (buoyancy-controlled or fixed at sea floor), while
the latter category of UWSNs are usually mobile since it is natural to imagine that the cost of deploying/recovering
fixed sensor nodes is typically forbidden for short-term time-critical applications. To summarize, we will focus on
three types of UWSNs: 1) Mobile UWSNs for long-term non-time critical applications (M-LT-UWSNs for short);
2) Static UWSNs for long-term non-time critical applications (S-LT-UWSNs for short); 3) Mobile UWSNs for
short-term time-critical applications (M-ST-UWSNs for short).
Obviously, different types of UWSNs have different communication requirements. We summarize the communi-
cation requirements for all three types of UWSNs in Table I.
Besides the UWSNs we discussed above, underwater networks also include sparse mobile AUV (autonomous
underwater vehicle) or UUV (unmanned underwater vehicle) networks, where vehicles/nodes can be spaced out by
several kilometers. This type of networks have their unique communication requirements, which are not the focus
of this paper.
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 4

TABLE I
C OMMUNICATION R EQUIREMENTS OF UWSN S

Requirements M-LT-UWSNs S-LT-UWSNs M-ST-UWSNs

Data Rate Various Various Various
Transmission Range Short (10m-1km) Short (10m-1km) Short (10m-1km)
Deployment Depth Shallow Water Shallow or Deep Shallow Water
Energy Efficiency Major Concern Major Concern Minor Concern
Antenna Size Small Small Small
Real-time Delivery Minor Concern Minor Concern Major Concern

III. F UNDAMENTALS OF P HYSICAL WAVES AS U NDERWATER C OMMUNICATION C ARRIERS

Understanding the first principles of each physical wave used in UWSN wireless communication is critically
important. In this section we layout the fundamental physical properties and critical issues for each of the acoustic,
EM, and optical wave propagations in underwater environments. We discuss each physical carrier’s advantages and
disadvantages towards efficient underwater wireless communication.

A. Acoustic Waves

Among the three types of waves, acoustic waves are used as the primary carrier for underwater wireless
communication systems due to the relatively low absorption in underwater environments. We start the discussion
with the physical fundamentals and the implications of using acoustic waves as the wireless communication carrier
in underwater environments.

1) Physical Properties: Acoustic waves has a number of propagation characteristics that are unique from other
waves, two of which are highlighted below.

Propagation velocity: The extremely slow propagation speed of sound through water is an important factor that
differentiates it from electromagnetic propagation. The speed of sound in water depends on the water properties of
temperature, salinity and pressure (directly related to the depth). A typical speed of sound in water near the ocean
surface is about 1520 m/s, which is more than 4 times faster than the speed of sound in air, but five orders of
magnitude smaller than the speed of light. The speed of sound in water increases with increasing water temperature,
increasing salinity and increasing depth. Most of the changes in sound speed in the surface ocean are due to the
changes in temperature. This is because the effect of salinity on sound speed is small and salinity changes in
the open ocean are small. Near shore and in estuaries, where the salinity varies greatly, salinity can have a more
significant effect on the speed of sound in water. As depth increases, the pressure of water has the largest effect
on the speed of sound.
Under most conditions the speed of sound in water is simple to understand. Sound will travel faster in warmer
water and slower in colder water. Approximately, the sound speed increases 4.0 m/s for water temperature arising
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 5

Fig. 1. A vertical profile of sound speed in seawater as the lump-sum function of depth (from
http://www.coexploration.org/bbsr/classroombats/html/sofarfig4.html).

1◦ C. When salinity increases 1 practical salinity unit (PSU), the sound speed in water increases 1.4 m/s. As the depth
of water (therefore also the pressure) increases 1 km, the sound speed increases roughly 17 m/s. It is noteworthy
to point out that the above assessments are only for rough quantitative or qualitative discussions, and the variations
in sound speed for a given property are not linear in general. The overall effect of sound speed in seawater is
illustrated in Fig. 1.
The slow propagation speed of sound impacts communication system performance and network protocol design
in a number of ways, as will be discussed in later sections on engineer countermeasures and networking challenges.

Absorption: During propagation, wave energy may be converted to other forms and absorbed by the medium. The
absorptive energy loss is directly controlled by the material imperfection for the type of physical wave propagating
through it. For acoustic waves, this material imperfection is the inelasticity, which converts the wave energy into
heat (while for EM waves the imperfection is the electric conductivity, as will be discussed in Section III-B.).
The absorptive loss for acoustic wave propagation is frequency-dependent, and can be expressed as eα(f )d , where
d is the propagation distance and α(f ) is the absorption coefficient at frequency f . For seawater, the absorption
coefficient at frequency f in kHz can be written as the sum of chemical relaxation processes and absorption from
pure water [5]:
A1 P1 f1 f 2 A2 P2 f2 f 2
α(f ) = + 2 + A3 P3 f 2 , (1)
f12 + f 2 f2 + f 2
where the first term on the right side is the contribution from boric acid, the second term is from the contribution of
magnesium sulphate, and the third term is from the contribution of pure water; A1 , A2 , and A3 are constants; the
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 6

pressure dependencies are given by parameters P1 , P2 and P3 ; and the relaxation frequencies f1 and f2 are for the
relaxation process in boric acid and magnesium sulphate, respectively. Fig. 2 shows the relative contribution from
the different sources of absorption as a function of frequency, and the variation in total absorption with frequency
for four different oceans [5].

Fig. 2. Absorption in generic seawater (left panel) and in different oceans (right panel) (from Ainslie and McColm, 1998 [5]).

2) Multipath: An acoustic wave can reach a certain point through multiple paths. In a shallow water environment,
where the transmission distance is larger than the water depth, wave reflections from the surface and the bottom
generate multiple arrivals of the same signal. In deep water applications, surface and bottom reflections may be
neglected. Due to the spatially varying sound speed, the wave refractions, however, can cause significant multipath
phenomena.
Assume that there are P distinct paths between the source and the receiver, and let τp denote the propagation
delay for the pth path. Further, we use D to denote the channel delay spread, defined as the time difference between
the first and the last arrivals of multipath propagation, i.e., D = τP −1 − τ0 . Due to the slow speed, the channel
delay spread from multipath propagation is large. For example, two physical arrivals that differ 15 meters in path
length leads to an arrival time difference of 10 ms (here we assume the propagation speed of sound is 1500 m/s).
Typical underwater channels have delay spread around 10 ms, but occasionally delay spread can be as large as 50
to 100 ms [6].
Large channel delay spread introduces time dispersion of a signal, which causes severe inter-symbol interference.
Consider a signalling scheme with bandwidth of B = 4 kHz. Each symbol interval is about Ts = 0.25 ms. In the
presence of a channel with delay spread of 10 ms, each symbol will affect the subsequent 10/0.25 = 40 symbols
due to waveform spreading. This brings grand challenges for efficient modulation and demodulation.

3) Path Loss: We now discuss the energy loss of channels.

For any propagation wave, there are three primary mechanisms for energy loss: (i) geometric spreading, (ii)
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 7

absorptive loss, and (iii) scattering loss. The absorptive loss for acoustic waves has been discussed in Section III-A1.
We next focus on geometric spreading and scattering loss.
Geometric spreading is the local power loss of a propagating acoustic wave due to energy conservation. When an
acoustic impulse propagates away from its source with longer and longer distance, the wave front occupies larger
and larger surface area. Hence, the wave energy in each unit surface (also called energy flow) becomes less and less.
For the spherical wave generated by a point source, the power loss caused by geometric spreading is proportional
to the square of the distance. On the other hand, the cylindrical waves generated by a very long line source, the
power loss caused by geometric spreading is proportional to the distance. For a practical underwater setting, the
geometric spreading is a hybrid of spherical and cylindrical spreading, with the power loss to be proportional to dβ ,
where β is between 1 (for cylindrical spreading) and 2 (for spherical spreading) [7]. Note that geometric spreading
is frequency-independent.
Scattering is a general physical process whereby one or more localized non-uniformities in the medium, such
as particles and bubbles, force some forms of wave radiation to deviate from a straight trajectory. It also includes
deviation of reflected radiation from the angle predicted by the law of reflection. This is especially relevant to
underwater channels. When the wind speed increases, the surface roughens and the effect of surface scattering
becomes evident [8]. Surface scattering introduces not only power loss, but also spreading in delay of each surface
bounce path (thus contributes to multipath phenomena as discussed in Section III-A2).

Now we are ready to formulate path loss. Still assume that there are P paths, and let ξp denote the scattering
loss, dp the propagation distance and τp the propagation delay of the pth path. Then the pass loss along the pth
path can be written as dβp eα(f )dp ξp , combining the effects of spreading loss, absorptive loss, and scattering loss.
Assuming that the channel is static within a certain time interval, the channel transfer function at frequency f can
be described as
P
X −1
1
H(f ) = q e−j2πf τp . (2)
p=0 dβp eα(f )dp ξp
We can then easily draw a conclusion that the overall channel attenuation is dependent not only on the distance, but
also on the frequency. Since α(f ) increase as f increases, high frequency waves will be considerably attenuated
within a short distance, while low frequency acoustic waves can travel very far. As a result, the bandwidth is
extremely limited for long-range applications, while for short-range applications, several tens of kHz bandwidth
could be available (a thorough study on the relationship between bandwidth (capacity) and distance is reported in
[9]). Therefore, acoustic waves are considered practical for efficient communication in underwater sensor networks,
where sensors are usually densely deployed.

4) Ambient Noise: Ambient noise is defined as “the noise associated with the background din emanating from a
myriad of unidentified sources. Its distinguishing features are that it is due to multiple sources, individual sources
are not identified, and no one source dominates the received field” [10]. The common sea-surface noise sources
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 8

include the surface-ship radiated noises, breaking waves associated with ensuing bubble production, and so on; and
the deep water noises mainly come from marine animals. Moreover, surface ships that cross ocean basins could
produce a general low frequency background traffic noise that may not in fact sound like coming from surface
shipping [11].
The level of underwater ambient noise may have large fluctuations upon a change in time, location or depth.
Nevertheless, it is still possible to sketch out a function describing the approximate magnitude range to characterize
underwater ambient noises in very general terms. Often pressure spectral density, defined as the mean squared
pressure of noise within a given frequency band divided by the bandwidth ∆f , is used. The unit of pressure
spectral density is pressure squared per Hertz. Fig. 3 plots the compiled spectral density from difference water
bodies over the world oceans. It should be noted that noise level is frequency-dependent. Thus, when selecting a
suitable frequency band for communication, besides path loss, noise should be also considered [8], [9].

10-2
Intensity Spectral Density (W/m /Hz)
2

10-6

10-10

10-14

101 102 103 104 105

Frequency (Hz)

Fig. 3. Noise intensity spectral densities for different environments in air and under water. Solid black: Interstate-5 in Seattle, 4 m off
center of right lane; Dashed black: Quiet city residence with distant traffic noise; Dotted black: Background noise at Hermit Basin, Grand
Canyon National Park; Solid gray: High shipping traffic and sea state 6 (strong gale); Dashed gray: Low shipping traffic and sea state 0
(glassy calm); Dotted gray: Under smooth sea ice in the Antarctic with no wind (from [12]).

Combining path loss and ambient noise, we may see the following effects on communication and networking:
For short-range acoustic communication, the level of ambient noise may be well below the desired signal. For long-
range or covert acoustic communication, the noise level would be a limiting factor for communication performance.
For networking, the most severe effect may be from some impulsive noises. The presence of this kind of noises
may cause highly dynamic link error rate or even link outage, which brings great challenges for networking design.

B. Electromagnetic Waves

The use of EM waves in radio frequency band has several advantages over acoustic waves, mainly on faster
velocity and high operating frequency (resulting in higher bandwidth). However, there are many limiting factors
when using EM waves in water. In this section, we will first discuss the fundamental physical behavior of EM field
in underwater environments. We then analyze the practicality of using EM for UWSNs
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 9

Due to the fact that EM field propagates very differently in freshwater and seawater, we describe EM in these
two types of media separately as follows.
EM in Freshwater. Freshwater is a low-loss medium. The propagation speed c can be expressed as [13]
1
c≈ √ , (3)
²µ
where ² is the dielectric permittivity, and µ is the magnetic permeability, whose value has no significant changes for
most non-magnetic media. The dielectric permittivity ² can be further expressed as the product of the permittivity
in air, ²0 (= 10−9 /(36π)), and the dimensionless relative permittivity, ²r (also known as the dielectric constant).
Since ²r for water (saline and fresh alike) is about 81, the speed of underwater EM waves is slowed down by only
a factor of 9 of the speed of light in free space. Clearly this speed is still much faster than that of underwater
acoustic waves, by more than 4 orders of magnitude, and it poses no problem in channel latency.
The absorption coefficient α for EM propagation in freshwater can be approximated as [13]
r
σ µ
α≈ , (4)
2 ²
where σ is the electric conductivity. Note that here the absorptive loss is essentially frequency-independent, and
EM waves can literally propagate through freshwater body. For example, ground penetrating radar (GPR) has
been successfully operated on the lake surface to map lake-bottom sediments. As such, using EM waves as the
communication carrier in freshwater environments appears very attractive. However, the key problem in using EM
waves for communication in freshwater underwater sensor networks is the antenna size. The big antenna size of
an EM transmitter (e.g., a couple of meters for a 50 MHz antenna) is unpractical for the dense deployment of
underwater sensor networks.
EM in Seawater. Seawater is a high-loss medium. The electric conductivity σ of seawater is about two orders
higher than that of freshwater. The higher conductivity in seawater is mainly due to the cumulative increase of total
dissolved solid (TDS) concentration in oceans, shown as the great salinity; the average salinity in seawater is about
34 parts per thousand (ppt). In highly conductive media, both the propagation velocity and the absorptive loss of
EM waves are functions of carrier frequency. The propagation speed of EM waves in seawater can be expressed
as [13] s
4πf
c≈ , (5)
µσ
while the absorption loss can be approximated as [13]
p
α≈ πf µσ. (6)

A plot of the velocity and the absorption coefficient versus frequency for EM waves in seawater is provided in
Fig. 4. Note that they are now frequency-dependent, approximately proportional to the square root of frequency.
This is the primary motivation for using lower frequency in highly conductive media. Seawater is a perfect example
of this type of media.
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 10

Fig. 4. Velocity and absorption versus frequency for EM waves in seawater.

Practicality of EM in Water For a given medium, the ratio of the electric conductivity and the dielectric permittivity,
σ/², referred to as transition frequency, defines the border of the behavior of an electromagnetic (EM) field in that
medium. If the frequency of an EM field is lower than the transition frequency, it behaves mostly like a diffusion
field; if the frequency is higher than the transition frequency, the EM field is mostly like a propagating wave. For
seawater, the conductivity σ is about 4 Siemens/meter, and the dielectric permittivity ² is 81 × 10−9 /(36π). These
values yield a transition frequency of about 4 × 36π × 109 /(2 × 81π) = 888 MHz. This means that if a carrier
working on the frequency of 10 MHz in seawater, which is much lower than seawater’s transition frequency, then
the EM field basically is not a wave anymore and it rather behaves like a diffusion field. On the other end of the
spectrum, if a carrier with frequency of 1 GHz is used, the EM field will mostly behave like a wave. However, due to
the high absorption of seawater (see Equation 6), the EM wave can hardly propagate. Therefore, EM communication
in seawater is literally unpractical when using classical approaches based on wave propagation.

In summary, the key limitation of using EM waves in freshwater is the big antenna size, and the critical problem
of using EM waves in seawater is the high attenuation. Thus, to make the use of EM waves practical for underwater
sensor network communication, more innovative approaches must be sought.

C. Optical Waves

Using optical waves for communication obviously has a big advantage in data rate, that can potentially exceed
1 Giga bps. However, there are a couple of disadvantages for optical communication in water. Firstly, optical
signals are rapidly absorbed in water. Secondly, optical scattering caused by suspending particles and planktons is
significant. Thirdly, high level of ambient light in the upper part of the water column is another adverse effect for
using optical communication.
Now let us constrain our discussion to the situation of using only monochromatic light in deep water (where
ambient light is usually not a major issue). Then optical scattering is the topic more pertinent to using optical waves
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 11

for communication. The scattering process of optical waves and the wavelength dependence of underwater optical
channels can be evaluated by the Mie scattering theory [14]. In contrast to Rayleigh scattering [14], which is valid
in the region where the wavelength is much larger than the size of the scattering particles, the Mie solution to the
scattering problem is rigorously valid for all possible ratios of particle diameter to wavelength.
According to the Mie theory, when the light wavelength is similar to the particle diameter, light interacts with the
particle over a cross-sectional area larger than the geometric cross section of the particle. The Mie theory provides
scattering cross section Csca , defined as the total energy scattered by a particle in all directions, as [14]
R 2π R π
Isca r2 sin φ dφ dθ
Csca = 0 0 , (7)
I0
where Isca is the scattered light intensity, I0 is the incident light intensity, and r is the radius of the particle.
The integration in (7) goes over the entire surface area of the sphere. When multi-scattering is predominant, i.e.,
when water has numerous suspending particles in a unit volume, the scattering cross section Csca is related to the
transmission of a light beam through multiple scatterers.
The attenuation due to optical scattering can be expressed as [14]:
dI
= −ζI, (8)
dx
where I is the light intensity, and ζ is the turbidity. Turbidity is a measure of the amount of cloudiness or haziness
in seawater caused by suspending particles. Turbidity provides an indication of the clarity of the seawater and is
measured using the nephelometric turbidity units (NTU). Seawater has a wide range turbidity, varying from tens to
several thousands of NTU [15]. Solving the above ordinary difference equation about multi-scattering light intensity
leads to a solution very similar to EM absorption:

Id = I0 e−ζd , (9)

where Id is the intensity at distance d through the medium with multiple scatterers, and I0 is the incident light
intensity. Obviously, the role of the turbidity ζ is exactly the same as the absorption coefficient α in wave absorption
loss. However, the physics is completely different: absorption is the power loss due to energy conversion to heat,
while scattering is the power loss due to energy diffraction to all directions.
The measure of contribution from individual scatterers to the total scattering is through turbidity. For the simplest
case of all scatterers possessing the same size, a simple relation exists for turbidity [14]:

ζ = N Csca , (10)

where N is the number of particles in unit volume, and Csca is the scattering cross section of an individual particle.
For more complicated case of multi-scattering, for example, if the sizes of the particles are not the same, the
turbidity will have a more complicated relationship with individual scattering particles as
Z
ζ = Csca (x)p(x)dx, (11)
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 12

where x is the particle diameter, Csca (x) is the scattering cross section for particles with diameter x; and p(x) is
the probability distribution function of particle size.
Based on the Mie scattering theory, the light intensity can be accurately estimated when using light as the carrier
for underwater sensor network communication. Apparently, the accurate knowledge of water turbidity is the first
requirement to estimate the range of communication with the use of Mie theory.
In short, in addition to the common issues of absorption loss and ambient “noise” from the environment as for
other waves, water turbidity plays an important role in deciding whether optical waves can be used as communication
carriers for underwater sensor networks.

D. Summary

For a more intuitive comprehension, we summarize the major characteristics of acoustic, electromagnetic and
optical carriers in Table II.
TABLE II
C OMPARISON OF ACOUSTIC , EM AND O PTICAL WAVES IN S EAWATER E NVIRONMENTS

Acoustic Electromagnetic Optical

Nominal speed (m/s) ∼ 1,500 ∼ 33,333,333 ∼ 33,333,333
Power Loss > 0.1 dB/m/Hz ∼ 28 dB/1km/100MHz ∝ turbidity
Bandwidth ∼ kHz ∼ MHz ∼ 10-150 MHz
Frequency band ∼ kHz ∼ MHz ∼ 1014 –1015 Hz
Antenna size ∼ 0.1 m ∼ 0.5 m ∼ 0.1 m
Effective range ∼ km ∼ 10 m ∼ 10-100 m

Apparently, each of the three physical wave fields physically has its own advantages and disadvantages for acting
as an underwater wireless communication carrier. The engineering ways taking the advantages and overcoming the
shortfalls of different carriers will be discussed in the next section.

IV. E NGINEERING C OUNTERMEASURES

In this section, we describe the engineering countermeasures that have been developed to address the physical
challenges for each wave used as the communication carrier in underwater sensor networks. These are physical
layer techniques to achieve point-to-point communication among sensor nodes.

A. Acoustic Communication (ACOMM)

Acoustic waves propagate well in seawater and can reach a far distance, as is the main reason why acoustic waves
are widely used in underwater communication. The main limitations and challenges of ACOMM are summarized
as follows.
First, acoustic communication is fundamentally bandwidth-limited. Frequencies up to 1 MHz have been tried in
field tests [6]. However, the usable frequency band depends on the transmission distance, as discussed in Section III.
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 13

Very-high-frequency bands (e.g., above 50 kHz) can be used only for short-range communication. For moderate
range communication, frequency range from 20 kHz to 50 kHz is often used. Low frequency waves (e.g., below
10 kHz) are effective for very long-range communication, e.g., in the order of tens of kilometers. The bandwidth
for ACOMM is typically in the order of kHz to tens of kHz, which is far inferior to that of radio communication.
How to utilize the limited bandwidth efficiently is one major objective for ACOMM, as amounts to increasing
the number of bits per second communicated per unit bandwidth (bits/sec/Hz), which is usually called bandwidth
efficiency.
Second, acoustic communication is severely interference-limited. Besides impulsive ambient noises, the major
source of interference is the self-interference induced by the time- and frequency-dispersive nature of the underwater
acoustic channel. On the one hand, the slow speed of acoustic waves and significant multipath phenomena cause very
large channel delay spread, which leads to severe inter-symbol interference due to the waveform time-dispersion
(also called time-spreading). On the other hand, in motion environments (such as platform motion and scattering
of the moving sea surface), the slow propagation speed of sound introduces large Doppler spread or shifts, which
causes severe interference among different frequency components of the signal (also referred to as frequency-
spreading). On the outset, large Doppler spread results in a reduction in the channel coherence time (the time
period when the channel can be viewed as static) or an apparent increase in the rate of channel fluctuation [8].
For example, consider v = 1.5 m/s and fc = 30 kHz, where v is the rate of change of the propagation path length
(e.g, the platform velocity), and fc is the carrier frequency. The Doppler shift frequency fd at frequency fc is
given by fd = v/cfc = 30 Hz, where c is the speed of sound in water. Further assume that a signal bandwidth
of 4 kHz is used, which results a symbol duration of T = 0.25 ms. The normalized Doppler per symbol time is
fd T = 0.75 · 10−2 . This implies that channel variation shall be accounted for on a symbol by symbol basis. Having
large delay and Doppler spreads at the same time entails a complex interference pattern that is hard to deal with.
In short, the objective of underwater acoustic communication is to overcome the performance limitations induced
by the highly dispersive channel, while at the same time improve the bandwidth efficiency as much as possible. We
next discuss various approaches that have been used in underwater acoustic communication.

Frequency shift keying (FSK). In FSK modulation, information bits are used to select the carrier frequencies of the
transmitted signal. The receiver compares the measured power at different frequencies to infer what has been sent.
Using only energy detector at the receiver, this scheme bypasses the need for channel estimation, and is thus robust
to channel variations. However, guard bands are needed to avoid the interference caused by frequency-spreading,
and guard interval is inserted between successive symbol transmissions for channel clearing to avoid the interference
caused by time-spreading. As a result, the data rate of FSK is very low. Frequency hopped (FH) FSK improves
the data rate as it does not need to wait the channel clearing corresponding to the previous symbol transmission
on a different frequency. However, due to the bandwidth expansion via frequency hopping, the overall bandwidth
efficiency remains low, typically much below 0.5 bits/sec/Hz.
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 14

Commercial modems such as those from Teledyne-Benthos [16] are based on FSK. The WHOI Micro-Modem
has two operating modes, with the low-power low-rate mode based on non-coherent FSK of 80bps over a bandwidth
of 4 kHz [17].

Direct sequence spread spectrum (DSSS). In DSSS modulation, a narrow band waveform of bandwidth W is
spread to a large bandwidth B before transmission. This is achieved by multiplying each symbol with a spreading
code of length B/W , and transmitting the resulting sequence at a high rate as allowed by bandwidth B . Multiple
arrivals at the receiver side can be separated via the de-spreading operation which suppresses the time-spreading
induced interference, thanks to the nice auto-correlation properties of the spreading sequence. Channel estimation
and tracking are needed if phase-coherent modulation such as phase-shift-keying (PSK) is used to map information
bits to symbols before spreading [18]. For noncoherent DSSS, information bits can be used to select different
spreading codes to be used, and the receiver compares the amplitudes of the outputs from different matched filters,
with each one matched to one choice of spreading code. This avoids the need for channel estimation and tracking.
DSSS is used in commercial modems such as those from LinkQuest [19], DSPCOMM [20], and Tritech [21].
Due to the spreading operation, the data rates are often in the order of hundreds of bps while using bandwidth of
several kHz, resulting in a bandwidth efficiency well below 0.5 bits/sec/Hz.

Single carrier phase-coherent modulation with adaptive channel equalization. One major step towards high rate
communication is the direct transmission of phase-coherent modulations, including phase shift keying (PSK) and
quadrature amplitude modulation (QAM) [22]. The channel introduces a great deal of inter-symbol interference (ISI)
due to multipath propagation. Advanced signal processing at the receiver side is used to suppress the interference;
this process is termed as channel equalization. Although widely used for slowly-varying multipath channels in radio
applications, channel equalization for fast-varying underwater channel is a big challenge. The canonical receiver in
[22] successfully combined a second-order phase-locked-loop to track channel phase variations with an adaptive
decision feedback equalizer to suppress the inter-symbol interference.
Without the guard interval insertion and the spreading operation, much higher data rates can be achieved with
single carrier phase-coherent modulation that those of FSK and DSSS. The WHOI Micro-Modem has a high-rate
mode that provides a variable rate from 300 to 5000 bit/s with a bandwidth of 4 kHz [17]. One concern about
single carrier transmission is that the receiver may be less robust as the parameters in the adaptive receiver need
to be fine-tuned depending on channel conditions.
When data symbols are transmitted at a higher rate, the same physical channel leads to more channel taps in
the discrete-time equivalent model. The complexity of time-domain equalization grows quickly as the number of
channel taps increases, which will eventually limit the rate increase for single-carrier phase-coherent transmission.
However, a frequency-domain equalization approach recently proposed in [23] may effectively deal with channels
with a large number of taps.
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 15

Multicarrier modulation. The idea of multicarrier modulation is to divide the available bandwidth into a large
number of overlapping subbands, so that the waveform duration for the symbol at each subband is long compared
to the multipath spread of the channel [24], [25]. Consequently, inter-symbol interference may be neglected in each
subband, greatly simplifying the receiver complexity of channel equalization. Precisely due to this advantage,
multicarrier modulation in the form of orthogonal frequency division multiplexing (OFDM) has prevailed in
recent broadband wireless radio applications. However, underwater channels entail large Doppler spread which
introduces significant interference among OFDM subcarriers. Lacking effective techniques to suppress the inter-
carrier interference (ICI), early attempts at applying OFDM to underwater environments had a very limited success.
Recently, there have been extensive investigations on underwater OFDM communication, including [26] on non-
coherent OFDM based on on-off-keying, [27] on a low-complexity adaptive OFDM receiver, and [28] on a pilot-tone
based block-by-block receiver. The block-by-block receiver does not rely on channel dependence across OFDM
blocks, and thus it is robust to fast channel variations across OFDM blocks [29], [30]. In contrast to single carrier
phase-coherent transmission, OFDM has one desirable property that one signal design can be easily scaled to fit
into different transmission bandwidths with negligible changes on the receiver [31]. With bandwidth varying from
3 kHz to 50 kHz, data rates from 1.5 kbps to 25 kbps after rate 1/2 coding and QPSK modulation are reported in
[31]. Further, with different bandwidths of 12 kHz, 25 kHz and 50 kHz, data rates of 12 kbps, 25 kbps, and 50
kbps after rate 1/2 coding and 16-QAM modulation are also achieved [31]. These recent studies demonstrate the
feasibility and flexibility of OFDM for underwater acoustic communication.

Multi-input multi-output techniques. A wireless system that employs multiple transmitters and multiple receivers
is referred to as a multiple-input multiple-output (MIMO) system. It has been shown that the channel capacity in a
scattering-rich environment increases linearly with min(Nt , Nr ), where Nt and Nr are the numbers of transmitters
and receivers, respectively [32], [33]. Such a drastic capacity increase does not incur penalty on precious power
and bandwidth resources, but rather it comes from the utilization of spatial dimension virtually creating parallel
data pipes. Hence, MIMO modulation is a promising technology to offer yet another fundamental advance on high
data rate underwater acoustic communication [34].
MIMO has been applied in both single carrier transmission and multicarrier transmission. For single carrier
transmission, existing adaptive channel equalization algorithms are leveraged to deal with MIMO channels [34],
[35]. The data rate increases substantially. For example, a 12kbps rate is achieved with 3 kHz bandwidth at the range
of 2 km, leading to a bandwidth efficiency of 4 bits/sec/Hz, using six transmitters and QPSK modulation [35]. Due
to OFDM’s unique strength in handling long dispersive channels with low equalization complexity, the combination
of MIMO and OFDM is another appealing solution for high data rate transmission but with low receiver complexity.
Reference [36] reports experimental results for a MIMO-OFDM system with two transmitters and four receivers,
where the data rate is 12 kbps with a 12 kHz bandwidth, leading to a bandwidth efficiency of 1 bits/sec/Hz, after
rate 1/2 coding and QPSK modulation, which doubles the efficiency of single antenna transmission in [30] when
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 16

using the same coding and modulation.

MIMO introduces additional interference among parallel data streams from different transmitters. Also, each
receiver has more channels to estimate, which requires more overhead spent on training symbols. For fast varying
underwater channels, the number of transmitters might not be large for best rate-and-performance tradeoff. In
addition to co-located antennas, distributed MIMO is also possible if clustered single-transmitter nodes could
cooperate [37]. Certainly, implementation of distributed MIMO needs to address challenging practical issues such
as node synchronization and cooperation.
In summary, there has been significant progress on ACOMM over recent years, in particular in the front of
multicarrier modulation and MIMO techniques. For short range communication with bandwidth in the order of
several tens of kHz and bandwidth efficiency in the order several bits per second per Hz, data rates up to 100 kbps
can be made available for underwater sensor networks.

B. Electromagnetic Communication (EMCOMM)

As discussed in Section III-B, the main challenge in using radio underwater is the severe attenuation due to the
conducting nature of seawater. As a result, EMCOMM works in the power-limited region.
Extremely low frequency radio signals have been used in military applications. Germans pioneered electromag-
netic communication in radio frequency for submarines during World War II, where the antenna was capable of
outputting up to 1 to 2 Mega-Watt (MW) of power [38]. An extremely low frequency (ELF) signal, typically around
80 Hz at much lower power, has been used to communicate with naval submarines globally today. This is possible
mainly because most of the transmission paths are through the atmosphere [38].
It was deemed impractical to use high frequency wave for communication purposes. However, theoretical analysis
and experiments show that radio waves within a frequency range 1 to 20MHz is able to propagate over distances
up to 100 m by using dipole radiation with transmission powers in the order of 100 W [39]. This will yield high
data rates beyond 1 Mbps which allows video images to be propagated at standard camera frame rates (25Hz) [40].
The antenna design in such case is very different from that of the antennas used for conventional service in the
atmosphere [38]–[40]. Instead of having direct contact with seawater, the metal transmitting and receiving aerials
are surrounded by waterproof electrically insulating materials [39], [40]. This way, an EM signal can be launched
from the transmitter into a body of seawater and picked up by a distant receiver.
In September 2006, the first commercial underwater radio-frequency (RF) modem in the world, model S1510,
was released by Wireless Fibre Systems [41]. Its data rate is 100 bps, and communication range is about several
tens of meters. In January 2007, a broadband underwater RF modem, model S5510, came into birth. It supports
1-10 Mbps within 1 meter range [41].
Due to the propagation property of EM waves, EMCOMM is an appealing choice only for very short range
applications. One example is the communication between autonomous underwater vehicles (AUVs) and base stations,
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 17

where the AUVs can move within the communication range of a base station to offload data and receive further
instructions [38].

C. Optical Communication (OCOMM) and Acousto-Optical Hybrid

As pointed out in Section III-C, water quality plays a key role in deciding whether optical waves can be used
for underwater communication. As a result, the applicability of OCOMM heavily depends on environments. Using
the same analogy for acoustic and EM waves, we say that OCOMM works in the environment-limited region.
So far, there are not many commercial activities on underwater optical communication, and no commercial
optical modems are available specifically for underwater. Recent interests in underwater sensor networks and sea
floor observatories have greatly stimulated the interest in short-range high-rate optical communication in water. For
example, in [42], an optical modem prototype is designed for deep sea floor observatories; in [43], a dual mode
(acoustic and optic) transceiver is used to assist robotic networks.
Lab testings has shown that very high data rates can be achieved within a short range. For example, a laboratory
experiment for underwater optical transmission achieves 1 Gbps rate over a 2-m path in a water pipe with up to 36
dB of extinction [44]. The source at 532 nm wavelength was derived from a 1064 nm continuous-wave laser diode
that was intensity modulated, amplified, and frequency-doubled in periodically poled lithium niobate. Measurements
were made over a range of extinction by the addition of Mg(OH)2 and Al(OH)3 suspension to the water path, and no
evidence of temporal pulse broadening was observed. Using Monte Carlo simulations over seawater paths of several
tens of meters indicates that optical communication data rates > 1 Gbps can be supported and are compatible with
high-capacity data transfer applications that require no physical contact [44].

Another interesting technique is the acousto-optical hybrid approach. In the linear regime of optical-acoustic
conversion, the laser beam incident at the air-water boundary is exponentially attenuated by the medium, creating an
array of thermo-acoustic sources relating to the heat energy and physical dimensions of the laser beam in water, thus
producing local temperature fluctuations that give rise to volume expansion and contraction. The volume fluctuations
in turn generate a propagating pressure wave with the acoustic signal characteristics of the laser modulation signal
[45]–[47]. Therefore, a number of acoustic signals, such as frequency modulated sweeps (also known as CHIRPs),
binary phase shift keyed (BPSK), quadrature phase shift keyed (QPSK), frequency shift keyed (FSK), and multi-
frequency shift keyed (MFSK) signals can be created for communication purposes.

D. Summary

We now summarize ACOMM , EMCOMM, and OCOMM in Table III. Combining Table II and Table III, we
can readily determine that EMCOMM and OCOMM are not suitable for underwater sensor networks (UWSNs)
with densely deployed nodes (referring to Table I for the communication requirements of UWSNs), or at least the
current techniques have not made EMCOMM and OCOMM practical for UWSNs. As for ACOMM, though it is
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 18

TABLE III
S UMMARY OF ACOMM, EMCOMM, AND OCOMM FOR UWSN IN SEAWATER ENVIRONMENTS

ACOMM EMCOMM OCOMM

Major hurdles bandwidth-limited, power-limited environment-limited
interference-limited
Data rate up to 100 kbps up to 10 Mbps up to 1Gbps
Antenna complexity medium high medium
Transmission range ∼ 50m-5km ∼ 1m-100m ∼ 1m-100m

applicable to UWSNs from the communication perspective, there are tremendous challenges in networking design,
which will be discussed in the next section.

V. N ETWORKING C HALLENGES FOR U NDERWATER ACOUSTIC S ENSOR N ETWORKS

In this section, we focus on the networking challenges for underwater acoustic sensor networks. Due to the
unique characteristics of underwater acoustic channels (long latency and low bandwidth) and the harsh underwater
environments (resulting in high channel dynamics), technology used in terrestrial radio networks could not be
applied to underwater acoustic networks. Next we discuss several typical networking problems in the design of
underwater sensor networks (UWSNs). We identify the design challenges and brief some recent solutions in the
literature.

A. Medium Access Control

Due to the dense deployment of sensors in UWSNs, we need to design an efficient medium access control
(MAC) protocol to coordinate the communication among sensors. This is a largely unexplored challenge in the
communication/networking community. On the one hand, there is no need for MAC protocols in existing small-
scale acoustic networks, since in such networks, sensors are sparsely separated from each other, and point-to-point
communication is sufficient. On the other hand, most existing MAC protocols in radio-based networks assume that
the signal propagation delay between neighbor nodes is negligible, as is significantly different from the scenario in
UWSNs, where the propagation delay of sound in water is five-magnitude higher than that of radio in air. Moreover,
the bandwidth capacities of acoustic channels are very low compared with those of RF channels. While ALOHA-
type of random access protocols used in satellite networks address the long delay issue to some extent, medium
access control handling both long propagation delay and low bandwidth is fairly uninvestigated. Furthermore, energy
efficiency of MAC protocols in satellite networks is usually not a major concern. In short, a viable MAC solution for
UWSNs should take long propagation delay, low available bandwidth, energy efficiency (for long-term applications)
and node mobility (for mobile UWSNs) into account.
So far, various approaches have been explored. Among the scheduling-based protocols (including time-division
multiple access (TDMA), frequency division multiple access (FDMA) and code division multiple access (CDMA)),
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 19

CDMA is considered a promising technique for underwater senor networks. In [48], a distributed CDMA scheme
is proposed. For contention-based protocols (where nodes compete for a shared channel, resulting in probabilistic
coordination), the applicability of random access methods and RTS/CTS-based approaches in underwater sensor
networks has been studied in [49]. Consistent with the conclusions drew in this paper, several protocols have been
proposed with different objectives. For example, in [50], Park et al. design a random access based MAC protocol,
called UWAN-MAC, for underwater sensor networks with very low and evenly distributed traffic. On the other hand,
coordination-based protocols, R-MAC [51] and T-Lohi [52] are suitable for dense underwater sensor networks with
high traffic rate.

B. Multi-hop Routing

Forwarding data from source nodes to command/control stations efficiently is very challenging in UWSNs,
especially in mobile UWSNs for long-term applications. In such networks, saving energy is a major concern. At
the same time, routing should be able to handle node mobility. This requirement makes most existing energy-
efficient routing protocols unsuitable for UWSNs. There are many routing protocols proposed for terrestrial sensor
networks, such as Directed Diffusion [53], and TTDD (Two-Tier Data Dissemination) [54]. These protocols are
mainly designed for stationary networks. They usually employ query flooding as a powerful method to discover data
delivery paths. In mobile UWSNs, however, most sensor nodes are mobile, and the “network topology” changes
very rapidly. The frequent maintenance and recovery of forwarding paths is very expensive in highly dynamic
networks, and even more expensive in dense 3-dimensional UWSNs.
Geographic routing is considered promising for mobile UWSNs. In [55], Xie et al. propose the first routing
protocol, called Vector-Based Forwarding (VBF), for mobile UWSNs. VBF is essentially a geographic routing
protocol [56]. It employs a novel concept of “routing vector”, which is defined as the vector connecting the source
to the sink. In VBF, the information of the routing vector is carried in each data packet. All nodes that are close
to the vector are qualified to forward data packets. In order to improve the robustness of VBF in sparse networks,
Nicolaou et al. [57] propose a hop-by-hop approach, called Hop-by-Hop Vector-Based Forwarding (HH-VBF). In
HH-VBF, the routing vector is not global any more. Instead, each forwarding node has a routing vector, which is
represented by the vector from the current node to the target.
Another critical issue challenges routing in UWSNs is the link outrage due to water turbulence, currents, obstacles
(e.g. ships), etc., as may cause intermittent network partitioning (that is, some nodes are disconnected from the
other nodes). There may be situations where no connected path exists at any given time between the source and
the destination. DTN (Delay/Disruption Tolerant Network) technique [58] shows promise in handling network
disruption. Guo et al. [59] have performed some initial study along this direction. In this work, an adaptive DTN
routing protocol is proposed for UWSNs. The adaptation is obtained by exploiting message redundancy and resource
reallocation in order to achieve different performance requirements.
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 20

C. Reliable Data Transfer

Reliable data transfer is important in UWSNs, especially for those aquatic exploration applications requiring
reliable information. There are typically two approaches to reliable data transfer: end-to-end and hop-by-hop. The
most common end-to-end solution TCP (Transmission Control Protocol). In UWSNs, due to the high and dynamic
channel error rates and the long propagation delay, TCP’s performance will be problematic. There are a number of
techniques that can be used to render TCP’s performance more efficient. However, the performance of these TCP
variants in UWSNs is yet to be investigated. Another type of approach for reliable data transfer is hop-by-hop.
The hop-to-hop approach is favored in wireless and error-prone networks, and is believed to be more suitable for
sensor networks [60]. There are a couple of reliable data transfer protocols proposed for terrestrial sensor networks,
such as PSFQ [60], RMST [61], and RBC [62]. These protocols mainly take hop-by-hop approach with ARQ. Due
to the long propagation delay of acoustic signals, conventional ARQ will cause very low channel utilization in
underwater environments. Thus, new approaches are desired for efficient reliable data transfer in UWSNs.
One possible direction to solve the reliable data transfer problem in UWSNs is to investigate coding schemes,
including erasure coding and network coding, which, though introducing additional computational and packet
overhead, can avoid retransmission delay and significantly enhance the network robustness. Xie et al. [63] propose
an approach called Segmented Data Reliable Transport (SDRT), which is essentially a hybrid approach of ARQ
and FEC. It adopts efficient erasure codes, transferring encoded packets block by block and hop by hop. Compared
with traditional reliable data transfer protocols used in underwater acoustic networks, SDRT can reduce the total
number of transmitted packets significantly, improve channel utilization, and simplify protocol management. In
[64], a network coding scheme is proposed for underwater sensor networks. This scheme carefully couples network
coding and multi-path routing for efficient error recovery.

D. Localization

Localization of mobile sensor nodes is indispensable for UWSNs. Some applications such as aquatic monitoring
demands high-precision localization, while other applications such as surveillance network requires a localization
solution that can scale to a large number of nodes. However, underwater acoustic propagation characteristics and
sensor mobility pose great challenges on high-precision and scalable localization solutions in that: i) underwater
acoustic channels are highly dispersive, and time delay of arrival (TDOA) estimation is hampered by dense multipath;
ii) acoustic signal does not travel on a straight path due to the stratification effect; iii) underwater acoustic channels
have extremely low bandwidth that renders any approach based on frequent message exchange not appealing;
iv) large scale sensor deployment prevents centralized solutions; and v) sensor mobility entails dynamic network
topology change.
To effectively handle the channel effects, high-precision localization usually involves advanced signal processing
algorithms. In [65], a depth-based approach is proposed to compensate the stratification effect for improved
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 21

underwater ranging. In the presence of dense multipath and fast channel variations, a multicarrier-signaling based
solution for precise timing and Doppler estimation is considered promising [66].
In the front of scalable localization, Zhou et al. [67] have proposed a hierarchical approach which divides
the whole localization process into two sub-processes: anchor node localization and ordinary node localization.
Many existing localization techniques for small scale networks (such as GIB [68], PARADIGM [69], and silent
positioning [70]) can be used in the former. For the ordinary node localization process, a distributed localization
scheme which novelly integrates a 3-dimensional Euclidean distance estimation method with a recursive location
estimation method has been developed. In a follow-up paper [71], a scalable localization approach with mobility
prediction has been investigated.

E. Summary

From the above discussions (though the problem list is far from complete), we can conclude that, although
acoustic waves are practical for underwater acoustic sensor networks from the physics and communication point
of view, a tremendous amount of work is demanded from the networking perspective.

VI. C ONCLUDING R EMARKS

Based on the discussion in previous sections, we have the following summary points.

• Up to date and extending to the near future, acoustic waves will be staying as the major carrier of wireless
communication in UWSNs. For acoustic wave carriers, apparently the key challenges are in communication
and networking.
• For electromagnetic radio wave carriers, the main shortcoming stays with the high absorption of EM waves
in water, especially in seawater. Though short-range wireless communication using EM waves in seawater has
seen certain breakthroughs, it will still be a long way to expand the approach to be used in UWSNs.
• Optical carriers will remain as to be used for some special applications. The major hurdle is that optical
communication in water is largely constrained by environments.

In short, this review article has analyzed the necessity of considering the physical fundamentals of an underwater
environment for a particular kind of physical wave to be used as the carrier of wireless communication among
nodes in an underwater sensor network. Acoustic wave remains the most robust and feasible carrier up to the date.

R EFERENCES

[1] I. F. Akyildiz, D. Pompili, and T. Melodia, “Underwater acoustic sensor networks: Research challenges,” Ad Hoc Networks, pp. 257–279,
2005.
[2] J.-H. Cui, J. Kong, M. Gerla, and S. Zhou, “Challenges: Building scalable mobile underwater wireless sensor networks for aquatic
applications,” IEEE Network, Special Issue on Wireless Sensor Networking, pp. 12–18, 2006.
[3] J. Heidemann, W. Ye, J. Wills, A. Syed, and Y. Li, “Research challenges and applications for underwater sensor networking,” in
Proceedings of the IEEE Wireless Communications and Networking Conference, 2006.
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 22

[4] J. Partan, J. Kurose, and B. N. Levine, “A survey of practical issues in underwater networks,” in Proceedings of ACM WUWNet’06,
2006.
[5] M. A. Ainslie and J. G.McColm, “A simplified formula for viscous and chemical absorption in sea water,” Journal of the Acoustical
Society of America, vol. 103, no. 3, pp. 1671–1672, 1998.
[6] D. B. Kilfoyle and A. B. Baggeroer, “The state of the art in underwater acoustic telemetry,” IEEE Journal of Oceanic Engineering,
vol. 25, no. 1, pp. 4–27, Jan. 2000.
[7] R. Urich, Principles of Underwater Sound. 3rd ed. New York: McGraw-Hill, 1983.
[8] J. Preisig, “Acoustic propagation considerations for underwater acoustic communications network development,” in Proceedings of First
ACM International Workshop on Underwater Networks (WUWNet), Los Angeles, CA, Sep. 2006.
[9] M. Stojanovic, “On the relationship between capacity and distance in an underwater acoustic communication channel,” in Proceedings
of First ACM International Workshop on Underwater Networks / MobiCom Conference, Los Angeles, CA, Sep. 2006.
[10] National Research Council (NRC), Ocean Noise and Marine Mammals. National Academies Press, 2003.
[11] G. M. Wenz, “Low-frequency deep water ambient noise along the pacific coast of the united states,” U.S. Navy J. Underwater Acoustics,
vol. 19, pp. 423–444, 1969.
[12] P. H. Dahl, J. H. Miller, D. H. Cato, and R. K. Andrew, “Underwater ambient noise,” Acoustics Today, vol. 3, no. 1, pp. 23–34, 2007.
[13] C. A. Balanis, Advanced Engineering Electromagnetics. John Wiley & Sons, New York, NY, 1989.
[14] M. Kerker, The Scattering of Light and other Electromagnetic Radiation, 3rd Ed. Academic Press, New York, NY, 1973.
[15] J. Zhang, Y. Liu, S. Gao, C. Li, F. Zhang, H. Zeng, and C. Ye, “Pilot testing of outside-in uf pretreatment prior to ro for high turbidity
seawater desalination,” Desalination, pp. 269–277, 2006.
[16] Benthos, Inc., “Fast and reliable access to undersea data,” http://www.benthos.com/pdf/Modems/ModemBrochure.pdf.
[17] L. Freitag, M. Grund, S. Singh, J. Partan, P. Koski, and K. Ball, “The WHOI Micro-Modem: An acoustic communications and navigation
system for multiple platforms,” in Proceeding of OCEANS, Washington DC, 2005.
[18] L. Freitag, M. Stojanovic, S. Singh, and M. Johnson, “Analysis of channel effects on direct-sequence and frequency-hopped spread-
spectrum acoustic communications,” IEEE Journal of Oceanic Engineering, vol. 26, no. 4, pp. 586–593, Oct. 2001.
[19] LinkQuest, Inc., “Underwater acoustic modems,” http://www.link-quest.com/html/uwm hr.pdf.
[20] DSPCOMM, “Underwater wireless modem,” http://www.dspcomm.com.
[21] Tritech International, “Tritech AM-300 acoustic modem,” http://www.auv.co.uk/products.
[22] M. Stojanovic, J. A. Catipovic, and J. G. Proakis, “Phase-coherent digital communications for underwater acoustic channels,” IEEE
Journal of Oceanic Engineering, vol. 19, no. 1, pp. 100–111, Jan. 1994.
[23] Y. R. Zheng, C. Xiao, T. C. Yang, and W.-B. Yang, “Frequency-domain channel estimation and equalization for single carrier underwater
acoustic communications,” in Proceedings of MTS/IEEE OCEANS conference, Vancouver, BC, Canada, Sep. 29 - Oct. 4 2007.
[24] J. A. C. Bingham, “Multicarrier modulation for data transmission: An idea whose time has come,” IEEE Communications Magazine,
pp. 5–14, May 1990.
[25] Z. Wang and G. B. Giannakis, “Wireless multicarrier communications: Where Fourier meets Shannon,” IEEE Signal Processing
Magazine, vol. 17, no. 3, pp. 29–48, May 2000.
[26] P. J. Gendron, “Orthogonal frequency division multiplexing with on-off-keying: Noncoherent performance bounds, receiver design and
experimental results,” U.S. Navy Journal of Underwater Acoustics, vol. 56, no. 2, pp. 267–300, Apr. 2006.
[27] M. Stojanovic, “Low complexity OFDM detector for underwater channels,” in Proceedings of MTS/IEEE OCEANS conference, Boston,
MA, Sep. 18-21, 2006.
[28] B. Li, S. Zhou, M. Stojanovic, L. Freitag, and P. Willett, “Multicarrier communication over underwater acoustic channels with
nonuniform Doppler shifts,” IEEE Journal of Oceanic Engineering, vol. 33, no. 2, Apr. 2008.
[29] B. Li, S. Zhou, M. Stojanovic, and L. Freitag, “Pilot-tone based ZP-OFDM demodulation for an underwater acoustic channel,” in
Proceedings of MTS/IEEE OCEANS conference, Boston, MA, Sep. 18-21, 2006.
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 23

[30] B. Li, S. Zhou, M. Stojanovic, L. Freitag, and P. Willett, “Non-uniform Doppler compensation for zero-padded OFDM over fast-varying
underwater acoustic channels,” in Proceedings of MTS/IEEE OCEANS conference, Aberdeen, Scotland, Jun. 18-21, 2007.
[31] B. Li, S. Zhou, J. Huang, and P. Willett, “Scalable OFDM design for underwater acoustic communications,” in Proceedings of Intl.
Conf. on ASSP, Las Vegas, NV, Mar. 3 – Apr. 4, 2008.
[32] G. J. Foschini and M. J. Gans, “On limits of wireless communication in a fading environment when using multiple antennas,” Wireless
Personal Communications, vol. 6, no. 3, pp. 311–335, Mar. 1998.
[33] I. E. Telatar, “Capacity of multi-antenna Gaussian channels,” Bell Laboratories Technical Memorandum, 1995.
[34] D. B. Kilfoyle, J. C. Preisig, and A. B. Baggeroer, “Spatial modulation experiments in the underwater acoustic channel,” IEEE Journal
of Oceanic Engineering, vol. 30, no. 2, pp. 406–415, Apr. 2005.
[35] S. Roy, T. M. Duman, V. McDonald, and J. G. Proakis, “High rate communication for underwater acoustic channels using multiple
transmitters and space-time coding: Receiver structures and experimental results,” IEEE Journal of Oceanic Engineering, vol. 32, no. 3,
pp. 663–688, Jul. 2007.
[36] B. Li, S. Zhou, M. Stojanovic, L. Freitag, J. Huang, and P. Willett, “MIMO-OFDM over an underwater acoustic channel,” in Proceedings
of MTS/IEEE OCEANS conference, Vancouver, BC, Canada, Sep. 29 - Oct. 4, 2007.
[37] M. Vajapeyam, U. Mitra, J. Preisig, and M. Stojanovic, “Distributed TR-STBC schemes for cooperative underwater acoustic
communications,” in Proceedings of OCEANs Asia Pacific, May 2006.
[38] T. Shelley, “Radio waves transmit information underwater,” An article in Eureka Magazine, Nov. 14, 2005.
[39] A. I. Al-Shamma’a, A. Shaw, and S. Saman, “Propagation of electromagnetic waves at MHz frequencies through seawater,” IEEE
Transactions on Antennas and Propagation, vol. 52, no. 11, pp. 2843–2849, Nov. 2004.
[40] J. Lucas, A. Al-Shamma’a, J. Seim, W. Loehr, G. Puchbauer, and D. McGregor, “Underwater communications using electromagnetic
waves (EMCOMMS),” in Proceedings of European Conference on Marine Science and Ocean Technology, Galway, Ireland, May 10–13,
2004.
[41] Wireless Fibre Systems, http://www.wirelessfibre.co.uk/.
[42] N. Farr, A. Chave, L. Freitag, J. Preisig, S. White, and D. Yoerger, “Optical modem technology for seafloor observatories,” in Proceedings
of OCEANS, 2005.
[43] I. Vasilescu, K. Kotay, D. Rus, M. Dunbabin, and P. Corke, “Data collection, storage, and retrieval with an underwater sensor network,”
in Proceedings of ACM Sensys, 2005.
[44] F. Hanson and S. Radic, “High bandwidth underwater optical communication,” Applied Optics, vol. 47, pp. 277–283, Jan. 2008.
[45] P. J. Westervelt and R. S. Larson, “Laser-excited broadside array,” Journal of the Acoustic Society of America, vol. 54, pp. 121–122,
1973.
[46] Y. H. Berthelot, “Thermoacoustic generation of narrow-band signals with high repetition rate pulsed lasers,” Journal of the Acoustic
Society of America, vol. 85, pp. 1173–1181, 1989.
[47] F. Blackmon, L. Estes, and G. Fain, “Linear opto-acoustic underwater communication,” Applied Optics, vol. 44, no. 18, pp. 3833–3845,
Jun. 2005.
[48] D. Pompili, T. Melodia, and I. F. Akyildiz, “A distributed CDMA medium access control in underwater acoustic sensor networks,” in
Proceedings of Med-Hoc-Net, Corfu, Greece, 2007.
[49] P. Xie and J.-H. Cui, “Exploring random access and handshaking techniques in large-scale underwater wireless acoustic sensor networks,”
in Proceedings of MTS/IEEE Oceans’06, Boston, MA, Sep. 2006.
[50] M. K. Park and V. Rodoplu, “Uwan-mac: An energy-efficient mac protocol for underwater acoustic wireless sensor networks,” IEEE
Journal of Oceanic Engineering, vol. 32, no. 3, pp. 710–720, Jul. 2007.
[51] P. Xie and J.-H. Cui, “R-MAC: An energy-efficient MAC protocol for underwater sensor networks,” in Proceedings of International
Conference on Wireless Algorithms, Systems, and Applications (WASA’07), Chicago, Illinois, USA, Aug. 1 - 3 2007.
[52] A. Syed, W. Ye, and J. Heidemann, “T-Lohi: A new class of MAC protocol for underwater acoustic sensor networks,” in Proceedings
of IEEE Infocom’08, 2008.
WILEY WCMC SPECIAL ISSUE ON UNDERWATER SENSOR NETWORKS (INVITED) 24

[53] C. Intanagonwiwat, R. Govindan, and D. Estrin, “Directed diffusion: A scalable and roust communication paradigm for sensor networks,”
in Proceedings of ACM International Conference on Mobile Computing and Networking (MOBICOM’00), Boston, Massachusetts, USA,
Aug. 2000.
[54] F. Ye, H. Luo, J. Cheng, S. Lu, and L. Zhang, “A two-tier data dissemination model for large-scale wireless sensor networks,” in
Proceedings of MOBICOM’02, Atlanta, Georgia, USA, Sep. 2002.
[55] P. Xie, J.-H. Cui, and L. Lao, “VBF: Vector-based forwarding protocol for underwater sensor networks,” in Proceedings of IFIP
Networking’06, Coimbra, Portugal, May 2006.
[56] D. Niculescu and B. Nath, “Trajectory-based forwarding and its application,” in Proceedings of ACM International Conference on
Mobile Computing and Networking (MOBICOM’03), San Diego, California, USA, Sep. 2003.
[57] N. C. Nicolaou, A. G. See, P. Xie, J.-H. Cui, and D. Maggiorini, “Improving the robustness of location-based routing for underwater
sensor networks,” in Proceedings of Oceans’07, Aberdeen, Scotland, Jun. 2007.
[58] K. Fall, “A delay tolerant networking architecture for challenged internets,” in Proceedings of ACM SIGCOMM 2003, Aug. 2003.
[59] Z. Guo, G. Colombi, B. Wang, J.-H. Cui, D. Maggiorini, and G. P. Rossi, “Adaptive routing in underwater delay/disruption tolerant sensor
networks,” in Proceedings of Fifth IEEE/IFIP Annual Conference on Wireless On demand Network Systems and Services (WONS’08),
Jan. 2008.
[60] C.-Y. Wan, A. T. Campbell, and L. Krishnamurthy, “PSFQ: A reliable transport protocol for wireless sensor networks,” in Proceedings
of WSNA’02, Atlanta, Georgia, USA, Sep. 2002.
[61] F. Stann and J. Heidemann, “RMST: Reliable data transport in sensor networks,” in Proceedings of 1st IEEE International Workshop
on Sensor Net Protocols and Applications(SNPA), Anchorage, Alaska, USA, May 2003.
[62] H. Zhang, A. Arora, Y. Choi, and M. G. Gouda, “Reliable bursty convergecast in wireless sensor networks,” in Proceedings of
Mobihoc’05, Urbana-Champaign, Illinios, USA, May 25-27 2005.
[63] P. Xie and J.-H. Cui, “An FEC-based reliable data transport protocol for underwater sensor networks,” in Proceedings of 16th
International Conference on Computer Communication and Networks (ICCCN’07), Honolulu, Hawaii, USA, Aug. 2007.
[64] Z. Guo, B. Wang, and J.-H. Cui, “Efficient error recovery using network coding in underwater sensor networks,” in Proceedings of
IFIP Networking’07, Atlanta, Georgia, USA, May 14-18, 2007.
[65] C. R. Berger, S. Zhou, P. Willett, and L. Liu, “Stratification effect compensation for improved underwater acoustic ranging,” IEEE
Transactions on Signal Processing, 2007.
[66] C. R. Berger, S. Zhou, Z. Tian, and P. Willett, “Precise timing for multiband OFDM in a UWB system,” in Proceedings of International
Conference on Ultra Wideband, Waltham, MA, Sep. 24-27, 2006.
[67] Z. Zhou, J. Cui, and S. Zhou, “Localization for large-scale underwater sensor networks,” in Proceedings of IFIP Networking’07, Atlanta,
Georgia, USA, May 14-18, 2007.
[68] C. Bechaz and H. Thomas, “GIB system: The underwater GPS solution,” in Proceedings of 5th Europe Conference on Underwater
Acoustics, May 2000.
[69] T. C. Austin, R. P. Stokey, and K. M. Sharp, “PARADIGM: A buoy-based system for AUV navigation and tracking,” in Proceedings
of MTS/IEEE OCEANS, vol. 2, 2000.
[70] X. Cheng, H. Shu, and Q. Liang, “A range-difference based self-positioning scheme for underwater acoustic sensor networks,” in
Proceedings of International Conference on Wireless Algorithms, Systems and Applications (WASA), Aug. 2007.
[71] Z. Zhou, J.-H. Cui, and A. Bagtzoglou, “Scalable localization with mobility prediction for underwater sensor networks,” in Proceedings
of IEEE INFOCOM’08, Mini-Conference, Phoenix, Arizona, USA, Apr. 13-18, 2008.